1. Sure, but it is also helpful to visualize these relationships through features like data lineage.

2. Agree, but perhaps not in the near future. A well defined semantic layer is likely a key part of the puzzle here and there is still a lot of manual pruning because metrics are ultimately a business decision and shift over time.

3. Great point. It's not LLM, but at Monte Carlo we utilize some advanced algorithms to help surface key assets that are frequently queried and have a large impact downstream on consumers.

4. Agreed. Some catalogs are already leveraging GPT to help with documentation.